Michael Brink, CTO of CASA Software.
As it stands in 2026, large-scale IT has reached an inflection point. Innovation has led to the development of many unprecedented capabilities but also introduced an equal amount of complexity and risk.
Before adding more technology this year, prudent organisations will likely shift their focus to identifying and addressing the blind spots that slow progress, especially as platforms multiply, environments scale and operational complexity increases.
The name of the game is trust.
Cyber security predictions for 2026 tell us that as lines blur between human error and machine intelligence, defence has never been more personal. These predictions note the biggest known vulnerability of 2026 will not be the tech but rather the trust. It highlights that people remain the easiest way in for malicious attackers, and AI just made social engineering nearly impossible to stop.
Let’s unpack some of these predictions.
Quantum computing is highlighted as a next-generation technology that uses the principles of quantum physics to solve complex problems far beyond the capability of traditional super computers.
Cyber threats targeting the cloud are said to be coming in faster than we can prepare for them, and old playbooks simply aren’t going to cut it.
Resilience won’t depend on having the most tools or protections, but rather on confidently knowing what (and who) can actually be trusted.
In 2026, resilience won’t depend on having the most tools or protections, but rather on confidently knowing what (and who) can actually be trusted.
People are the key to unlocking institutional secrets or infiltrating a company’s technological assets, especially when they are willingly, or unwittingly, coerced into doing so. Also, geopolitical pressures might provoke threat actors to initiate disruptive or aggravating cyber attacks on their advisories.
It is also confirmed that agentic AI − autonomous systems capable of accomplishing a specific goal with limited human supervision − will bolster the capabilities of attackers.
Cyber defence analysts and practitioners across the globe are urging corporations to prepare roadmaps for post-quantum protection − algorithms that can defend against attacks from quantum computers. This report says the industry is aligned on the danger and understands that because the migration takes years, the time to start is now.
It is widely reported that attackers have started targeting cloud platforms with less focus on conventional networks.
Visibility, automation and AI are rewriting the rules of network operations. 2026 is predicted to see further evolution of network observability practices, ensuring smarter, faster and more autonomous infrastructure.
AI is emerging as a stress test for existing cloud strategies. Even at the early stages of adoption, expectations around governance, cost transparency and data protection are already shaping infrastructure and application decisions. These pressures are exposing architectural gaps that may have been manageable for traditional workloads but become challenging when AI is introduced.
A common blind spot is reported to be that of treating AI as a standalone initiative rather than as a shared platform capability. In 2026, AI will move beyond isolated pilots and discrete services to become embedded across enterprise platforms, supporting application development, infrastructure operations and data processing workflows.
However, AI will only be as powerful as the network that supports it and less than 50% of existing networks are ready for AI, according to Dimensional Research.
Increased complexity, reliance on third-parties and the limitations of legacy NetOps tools − also referred to as traditional network management systems− are characterised by their focus on hardware-centric, manual processes designed for a more static IT environment.
These tools, which arose to manage specific, fragmented network tasks, now often hinder modern, cloud-native and agile network operations.
In 2026, the definition of network observability success will be measured by the ability to see, predict and explain the network’s current state holistically, incorporating the edge and cloud as well as the internet. It will no longer just focus on uptime or throughput.
Value of mainframe data for AI projects
Mainframes continue to run crucial commercial applications and will remain a cornerstone and strategic investment for large enterprises, and more recently expanded to include use cases for a rapidly growing number of generative AI and agentic AI projects.
According to Intellyx, c offer high-performance CPUs that support distributed virtual machines and native LLM training applications, thereby delivering new and significant value to any business investing in generative AI projects. Complexity and traditional approaches fuel the need for modern scalable observability solutions in 2026.
Tool sprawl and the complexities attached to managing the human resources responsible for each tool have become increasingly laborious and costly. Expert practitioners are difficult to recruit and retain, and siloed team structures incentivise process barriers and poor communications, especially where multiple tools from multiple vendors are part of the stack.
A new more progressive trend is emerging that promotes tool and agent consolidation, thereby unhinging agent cost of ownership and changing to best-of-suite tooling. AI is poised to become a powerful standard layer of this architecture, thereby creating more consistent outcomes.
In my next article on 2026 predictions, I will expand on securing the software supply chain without slowing down development.
